ProjectDescriptionSourceforge Page CodeDownload & InstallationBrowse (SVN) Coding Guidelines Tutorialslibeblearnlibidx DemosMNIST: Digit RecognitionNORB: Object Recognition FaceNet: Face Recognition Documentationlibeblearnlibidx |
DownloadThe project is still on pre-release version, so there is no pre-packed version of the library. Still, you are welcome to visit the sourceforge page and browse the SVN repository, where the project is updated very regularly at the moment.
You can get the current source tree with the following command: InstallationUbuntu
Required external packages (Non-Proprietary)all packages: libeblearn, libidx, tool libraries, tester and demos$
sudo apt-get install subversion cmake g++
libatlas-base-dev
libboost-filesystem-dev
imagemagick
libboost-regex-dev
libqt4-core libqt4-dev libqt4-gui
libcppunit-dev
libeblearn and libidx only (tools, testers and some demos will not compile)$
sudo apt-get install cmake g++
libatlas-base-dev imagemagick
for developers only (tools for bug-free development)$
sudo apt-get install valgrind electric-fence
Compilation$ ./configure
Execution
$ cd bin
MAC OS X
(by George Williams from
http://movement.nyu.edu/internal/Main/EblearnMacosNotes) Required external packages (Non-Proprietary)Install cmakeo I installed from the src tarball (cmake-2.6.2.tar.gz) which you can get here at http://www.cmake.org/cmake/resources/software.html.o Consult the README.txt in the top level directory and note that the steps to compile and install are "./bootstrap; make; sudo make install". Install libatlaso I installed from the src tarball (Atlas-C++-0.6.1.tar.gz) which you can get here at http://sourceforge.net/projects/worldforge/o The standard "configure/make/sudo make install" will do. Install cppunito I installed from the src tarball (cppunit-1.12.1.tar.gz) which you can get here at http://apps.sourceforge.net/mediawiki/cppunit/.o The standard "configure/make/sudo make install" will do. Install mnist databaseso You get these here http://yann.lecun.com/exdb/mnist/.o You will need to unpack the data o You will need the full path to the data directory later on Install boosto You should be able to install from the most recent boost version. I found a "macports" version, which is much easier to install.o Macports install - If you don't have macports, use the tarball instructions here: http://www.macports.org/install.php. I installed from the tarball MacPorts-1.7.0.tar.gz. o Macports boost - cd into the macports bin directory and type ">> sudo ./ports install boost". This may take a while. Download and adjust eblearn buildo Download from the subversion repository: ">> svn export https://eblearn.svn.sourceforge.net/svnroot/eblearn/trunk eblearn"o Modify the build to find boost headers. Locate the file "eblearn/scripts/FindBoost.cmake" and insert the line at the top of the file "SET(BOOST_INCLUDE_DIR /opt/local/include)". Assuming you did not change the default install directory of macports, then /opt/local/include will contain the macports versions of packages. Compilationo At this point, you can use the standard instructions to build and configure eblearno cd into the toplevel eblearn directory, and type ">>./clean.sh" (just in case you did a previous build with this src tree.) o Then type ">> ./build.sh". o Then type ">>./configure.sh". You'll need the full path to your local mnist data dir. Executiono cd into "bin" and type "./eblearn_tester".o all the implemented unit tests should pass |